Create app.py
Browse files
app.py
ADDED
|
@@ -0,0 +1,51 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import gradio as gr
|
| 2 |
+
import torch
|
| 3 |
+
from torchvision import transforms
|
| 4 |
+
from PIL import Image
|
| 5 |
+
import numpy as np
|
| 6 |
+
from unet_model import UNet
|
| 7 |
+
from huggingface_hub import hf_hub_download
|
| 8 |
+
|
| 9 |
+
# Load trained model weights from Hugging Face Hub
|
| 10 |
+
weights_path = hf_hub_download(
|
| 11 |
+
repo_id="Vizuara/unet-crack-segmentation", # ensure this matches your repo
|
| 12 |
+
filename="unet_weights_v2.pth" # make sure this file exists in repo
|
| 13 |
+
)
|
| 14 |
+
|
| 15 |
+
# Initialize and load model
|
| 16 |
+
model = UNet()
|
| 17 |
+
model.load_state_dict(torch.load(weights_path, map_location="cpu"))
|
| 18 |
+
model.eval()
|
| 19 |
+
|
| 20 |
+
# Preprocessing: same as training
|
| 21 |
+
IMG_HEIGHT, IMG_WIDTH = 128, 128
|
| 22 |
+
transform = transforms.Compose([
|
| 23 |
+
transforms.Resize((IMG_HEIGHT, IMG_WIDTH)),
|
| 24 |
+
transforms.ToTensor()
|
| 25 |
+
])
|
| 26 |
+
|
| 27 |
+
def predict(image):
|
| 28 |
+
orig_w, orig_h = image.size # original size of uploaded image
|
| 29 |
+
img = transform(image).unsqueeze(0) # (1,3,128,128)
|
| 30 |
+
with torch.no_grad():
|
| 31 |
+
pred = model(img)
|
| 32 |
+
|
| 33 |
+
mask = pred.squeeze(0).squeeze(0).cpu().numpy()
|
| 34 |
+
mask = (mask * 255).astype(np.uint8) # grayscale mask
|
| 35 |
+
|
| 36 |
+
# Resize back to original size
|
| 37 |
+
mask_img = Image.fromarray(mask).resize((orig_w, orig_h), Image.NEAREST)
|
| 38 |
+
return mask_img
|
| 39 |
+
|
| 40 |
+
|
| 41 |
+
# Gradio interface
|
| 42 |
+
demo = gr.Interface(
|
| 43 |
+
fn=predict,
|
| 44 |
+
inputs=gr.Image(type="pil"),
|
| 45 |
+
outputs=gr.Image(type="pil"),
|
| 46 |
+
title="UNet Crack Segmentation",
|
| 47 |
+
description="Upload a concrete surface image to get predicted crack mask"
|
| 48 |
+
)
|
| 49 |
+
|
| 50 |
+
if __name__ == "__main__":
|
| 51 |
+
demo.launch()
|